Working papers and full-length research from our Assessment Science team. Predictive validity, adverse impact, schools engagement effectiveness, AI-on-both-sides — all benchmarked on our 2.4M-application dataset, all openly published with full methodology. No NDA, no gate, no email harvesting.
A 42-page structural look at why early careers screening is failing UK employers, and the behavioural alternatives that actually predict first-year performance. Benchmarked on 2.4M applications across seven sectors and twelve years of two-sided data.
How TalentAssess scores correlate with first-year line-manager ratings across 110 production cohorts. Disaggregated by sector, role family and protected characteristic. 4-fifths rule satisfied.
The 4-fifths rule applied to early-careers behavioural assessment. How we test it pre-launch, monitor it in production, and what we do when a cohort flags. Audit-ready by design.
David examines why selecting on prior advantage masks real potential — and what the evidence says about building something better. Heavy on first principles, light on jargon.
How early talent teams are measuring real impact when traditional engagement metrics flatter — and what replaces them. Co-authored with Tom Reeves and the Schools team.
71% of UK candidates now use AI to write applications. 89% of employers use AI to screen them. We measured what happens to selection quality on both sides — and what to do about it.
Tracking 18,000+ early-career hires across financial services, professional services and engineering. First-gen hires from behavioural-evidence pipelines outperform CV-screened cohorts on 24-month retention by 18 percentage points.
Filed with HMT and IfATE in advance of the 2026 review. Joint paper with the ISE. The headline: levy uptake is up; quality is uneven; routeway design matters more than levy structure.
The five-criterion standard we use to decide whether a piece of candidate signal is worth scoring. Useful for procurement teams writing RFPs, and for vendors trying to answer them honestly.
800+ schools, 220+ in opportunity areas. Which engagement formats correlate with progression to apprenticeship, university, and direct employment — and which don’t.
The research above is openly published, fully methodologically transparent, and peer-reviewed by an external panel of five chartered business psychologists before release. You don’t need an NDA, an account, or a commercial relationship to read or cite anything we put out — if it’s on this page, it’s public.
If we’ve published a number, we’ve already had it checked.
Procurement teams writing social-value bids, journalists fact-checking a quote, in-house researchers replicating our methodology, students writing dissertations: take what you need. Attribution is appreciated, not required.
If you spot a methodological gap, an analytical error, or a place where the data doesn’t support the conclusion, we’d like to know — research@thetalentpeople.co.uk.
For candidates →